JEC COMPOSITES MAGAZINE - Issue #106 - July 2016 - 38

process

measurement [1, 5]. This data can serve
as input for controlling the differential
pressure ∆p=pautoclave-presin by adjusting
the resin pressure, for example. The
developed manufacturing process is a
single-line injection (SLI) process with
a cure temperature of 180°C and an
autoclave pressure of 0.6 MPa, which
was successfully used for composite ribs
[1] earlier.
Dry-fibre technologies make it possible
to use draping for more complex part
geometries that cannot be produced
using conventional prepreg-specific
manufacturing technologies such as
automated tape laying (ATL) and
automated fibre placement (AFP). The
Tenax® Dry Reinforcements HTS45
dry-fibre material, applied for the
-stiffeners, has a multi-material architecture made up of a +45° composite ply,
an interleave toughener, a -45° composite ply, another interleave toughener followed by a powder binder surface layer.
It has an areal weight of approximately
400 g/m2. The laminate stacks used for
the stiffeners are preformed in a vacuum-based preforming step, utilizing a
membrane technique. Aerospace gusset
fillers (noodles) are used at the junction
areas of the C-profiles and covers. This
is important in order to avoid neat resin
agglomeration in the filler area, which
is critical from a stress point of view.
Moreover, neat resin areas often lead to
cracks mainly induced by the considerable resin cure shrinkage.
Adequate resin flow control is mandatory when manufacturing aerospace-quality parts using an injection process.

Fig. 2: Manufactured CFRP component, segmented frame and final component developed by
Embraer, Toho Tenax and DLR

Fig. 3: Normalized PID predictions and compensation measures derived through the P-approach

Therefore, simulation techniques are
applied to assess resin flow and to define
robust injection strategies. For the
structure at hand, the injection concept
was designed using RTM-Worx simulations and in-house experience. Figure
2 gives an overview of the fabrication
steps, showing the manufactured composite structure on the left, the trimmed
interstage product in the middle and the
final part on the right.

Straightforward prediction-based
tool compensation
Process-induced distortions (PID)
occur inevitably when manufacturing
composite structures [3]. Distortions
arise from the material's anisotropy.
Reliable compensation strategies are
essential in order to fabricate nominal-shape parts. As the compensation
task often pops up when the part design
and stress analyses are finalized, it represents an additional work step within the
tool design process. Therefore, methods
with an excellent trade-off between prediction effort and prediction accuracy
are particularly demanded by composite
manufacturers. For the stiffener profile
at hand, the so called P-approach was
applied [2, 4]. This approach pursues a
semi-numerical strategy that focuses on
predicting process-induced distortions
of composite structures. In contrast to
state-of-the-art physically-based process
simulation approaches, which demand
costly material parameter characterization of fibres and resin as well as
complex numerical modelling, the P-approach utilizes conventional engineer-

38 jec composites magazine / No106 July 2016

ing constants and a single additional
parameter capturing distortion. This dimensionless parameter is derived from
the distortion of a manufactured reference specimen by an analytical transfer.
Even available data from the literature
or previous experience can be used for
a good qualitative estimation. The P-approach can be applied to conventional
solid and shell elements [4]. In the present case of the -stiffener, an efficient
2D cross-sectional model was used, as
the part's cross-section does not change
along the part length. The manufactured
cross-section is a junction of two C-profiles covered with two flat laminates
on either side. These covers affect the
overall distortion mode, which makes
a simple experience-based estimation
of process-induced distortions impossible. Therefore, the application of the
P-approach was mandatory to derive
the intended compensation measures.
As the web is not perpendicular to the
chords, asymmetric compensation
measures are derived for both the long
and the short legs of the profile. Figure
3 shows the normalized distortions of
the cross-section and the corresponding
compensated part shape.
Typically, the calculated process-induced distortions of a composite structure are inverted and fed back into the
nominal CAD model, whose geometry
is updated accordingly. However, in
many cases this complicates the part geometry significantly, as for example flat
areas become slightly double curved.
This significantly complicates the tool

Table of Contents for the Digital Edition of JEC COMPOSITES MAGAZINE - Issue #106 - July 2016